Gallup Tracking Poll Thread [Obama vs McCain]

[Alcon]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

[J. J.]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

There may have been a higher than normal sample for Obama midweek?

[they don't love you like i love you]

Three words certain people need to remember:

MARGIN OF ERROR

[Vsanto5]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

There may have been a higher than normal sample for Obama midweek?

Yesterday's Rasmussen poll seems to be the opposite.  If that is the case than McCain is in serious trouble.  It always seems the Rasmussen and Gallup polls are opposite of each other since earlier in the summer Rassmussen had Obama ahead and Gallup had the race closer, now Rassmussen has the race closer and Gallup has Obama outside the margin of error or within a reasonable prediction.

[Alcon]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

There may have been a higher than normal sample for Obama midweek?

Very possible, but again:  Consider rounding.  Consider margin of error.  We're seeing movement here at a very low confidence rate.  People trust polls too literally, and then assail them too much because of it.

[Sam Spade]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

There may have been a higher than normal sample for Obama midweek?

Very possible, but again:  Consider rounding.  Consider margin of error.  We're seeing movement here at a very low confidence rate.  People trust polls too literally, and then assail them too much because of it.

In other words, we need to look at the forest, not the trees...  Really? 

[Umengus]

We are still in summer guys...

[J. J.]

Hmm, I must say I am baffled how his lead didn't grow with Saturday polling included.

Because, for the second time, you are ignoring natural volatility/bad sampling, and assuming that this is a flat-lined race with Obama spikes on weekends.  Also, consider rounding; it's possible, maybe likely (tendency toward the mean) that the previous result was a "weak" +5.

There may have been a higher than normal sample for Obama midweek?

Yesterday's Rasmussen poll seems to be the opposite.  If that is the case than McCain is in serious trouble.  It always seems the Rasmussen and Gallup polls are opposite of each other since earlier in the summer Rassmussen had Obama ahead and Gallup had the race closer, now Rassmussen has the race closer and Gallup has Obama outside the margin of error or within a reasonable prediction.

The weekend have tended to be higher for Obama in Gallup, and not in Rasmussen.  I was actually thinking that there was something in the sample method that might account for it.  Now, with this week probably not, but I'm wondering if last weeks sample might have had  single skewed  sample (or, conversely, this one has a skewed sample for McCain).

[agcatter]

Rasmussen uses very hard party weighting (currently D + 9).  That means less variance on weekends.  Even when less Republicans are polled on weekends (which is normal) the Republican numbers are counted at 31.5% of the sample and the Democrats are weighted at 40.5% regardless of the number of Dems in the sample.

Rasmussen adjusts the weight of each party he uses based on a constantly rolling 3 month sample of 15,000 adults (5,000 per month) asking their Party ID.  Based on that he had it at D + 10 in July and tweeked it to D + 9 August 1st.

Interestingly his July sample of 5000 showed D + 7.8.  If he gets something around that number for his 5000 August sample, we can probably expect his D + 9 to decline a half a point or so beginning September 1st.  Or it could go up if his August party ID sample comes up more Democratic.

Gallup varies more on weekends because they don't use the hard party weighting and are reliant on whatever number of Republicans or Democrats they happen to pick up on a given night.  There will thus be more fluctuation especially on the weekends.

Rasmussen used the  method he's using now in 2004 and absolutely nailed it.

Of course, Rasmussen is using likely voters as opposed to registered in Gallup.

[??????????]

We are still in summer guys...

Of course. I'd personally say this race is a dead tie.

[agcatter]

I'd say it is Obama + 1.5 at present and definitely not the Obama + 6 or 7 you see in Pew and CBS/NYT, Newsweek etc.

[Sam Spade]

The Democratic margin in party ID will continue to decline over August and September - on that I will place money.

My question is - how is Gallup's tracking poll conducted, because presumably, it's not conducted like the normal Gallup poll is, otherwise we'd see the huge bumps - Alcon?

[Democratic Hawk]

Monday, August 11, 2008

Obama - 47% (+1)
McCain - 42% (-1)

The pattern of Americans' support for the two presumptive presidential nominees has shown little substantive change so far in August. There have been eight Gallup Poll Daily tracking reports based on interviewing conducted entirely in the month of August, and in each of these Obama has received either 46% or 47% of the support of registered voters. McCain has received between 42% and 44% support. These estimates of support for the two candidates show no significant change, a finding that is not surprising given the attention being given to the Olympics by voters, the fact that Obama is on vacation, and the general lack of major news generating events from the presidential campaign front.

The dramatic news of the sudden war which erupted in the Republic of Georgia has certainly been prominently displayed in newspapers, on television, and on Internet news sites, but the probability that these events will significantly change the U.S. presidential race is most likely low.

[RJ]

I really wish people would just give up on this "weekend polling" crap. I've lost track if the polls are supposed to be more in favor of which candidate and on what days. When the polls tighten/open up, everyone tries to justify why it's where it is with the equivalent of specious reasoning.

[J. J.]

I really wish people would just give up on this "weekend polling" crap. I've lost track if the polls are supposed to be more in favor of which candidate and on what days. When the polls tighten/open up, everyone tries to justify why it's where it is with the equivalent of specious reasoning.

But yet it is usually there, or a midweek bounce for McCain, whichever way you want to measure it.  I'll add that it's usually there in Gallup, and not Rasmussen.

I think it might have more to do with how Gallup is sampling than any movement in the race.

[RJ]

But yet it is usually there, or a midweek bounce for McCain, whichever way you want to measure it.  I'll add that it's usually there in Gallup, and not Rasmussen.

Here are the reaults from the last 15 days of the Gallup tracking poll:

Mon 7/28:      Obama +8
Tue 7/29:       Obama +6
Wed 7/30:      Obama +4
Thurs: 7/31:   Obama +1
Fri 8/1:           Tie
Sat 8/2:           Tie
Sun 8/3:          Obama +1
Mon 8/4:         Obama +3
Tue 8/5:          Obama +4
Wed 8/6:         Obama +2
Thurs: 8/7:      Obama +3
Fri 8/8:            Obama +3
Sat 8/9:            Obama +5
Sun 8/10:         Obama +3
Mon 8/11:        Obama +5

I'm just curious how you've interpreted this "bounce" from the data. Obama's lead may have seemed great Mon July 28/Tue July 29(which incidentally is the week after he got back from his Europe/Mideast trip) when compared to the later part of that particular week but it hovered around the same and if anything grew in the later part of last week when compared to mid or early week.

[Gustaf]

I believe the idea is that the trend has been there for more than 15 days. It obviously didn't turn out that way for the last week and you can argue away the week before that with Obama's Middle East trip and there are your two weeks. You would have to look further back to test the hypothesis.

[Democratic Hawk]

Tuesday, August 12, 2008

Obama - 47% (nc)
McCain - 42% (nc)

[Sam Spade]

There may actually be some slight movement towards Obama - but it's still too early to tell, yet.

[Lief 🗽]

I don't see how there can be movement when nothing is happening...

[Verily]

I don't see how there can be movement when nothing is happening...

I think Sam thinks people drift toward Obama when no one is paying attention. May or may not be true.

[Sam Spade]

I don't see how there can be movement when nothing is happening...

I think Sam thinks people drift toward Obama when no one is paying attention. May or may not be true.

Nope.    Just reading the broader week-to-week movements and when something starts to appear over the period of a week, it could mean that something is occurring.  I want to see the results tomorrow and Thursday before saying anything definite, however.

Factually though, when no one is paying attention, Democrats do poll better, because more of their base voters tend to be tuned in.  But that is not the present environment.

[Tender Branson]

Wednesday - August 13, 2008

Obama - 48% (+1)
McCain - 42% (nc)

[they don't love you like i love you]

By mid week, if there is no Obama VP announcement, McCain leads.

Ouch. Not even the next mid-week.

[Keystone Phil]

By mid week, if there is no Obama VP announcement, McCain leads.

Ouch. Not even the next mid-week.

Ok, dude, get over it/yourself.